RAG & Vector Search · Established · Intermediate

Hybrid Vector Keyword Search

Also known as: Hybrid Search, Dense-Sparse Search

Search architecture that combines dense vector embeddings with sparse BM25 keyword matching using reciprocal rank fusion.

What Hybrid Vector Keyword Search is

Hybrid Vector Keyword Search bridges the gap between semantic understanding and exact keyword matching (e.g., part numbers, names).

How it works

Runs parallel vector index queries and BM25 inverted index lookups, merging candidate ranks with weighted scoring.

Why it matters

Standard pattern in production vector databases for mission-critical enterprise search.

Common uses